Download Chapter 2 Discrete Random Variables

Survey
yes no Was this document useful for you?
   Thank you for your participation!

* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project

Document related concepts

Probability wikipedia , lookup

Statistics wikipedia , lookup

Ars Conjectandi wikipedia , lookup

Randomness wikipedia , lookup

Transcript
Chapter 2
Discrete Random Variables
2.1
Preliminaries
Definition 4 A random variable is a mapping (a function) from the sample space into real numbers.
• We can define an arbitrary number of different random variables on
the same sample space.
Ex: Toss a fair 6-sided die. Let the random variable X take on the value
1 if the outcome is 6, and 0 otherwise. Let the random variable Y be equal
to the outcome of the die. Illustrate the mappings from the sample space
associated with X and Y . (Note that {X = 1} = {outcome is 6} = A,
and {X = 0} = Ac .)
28
METU EE230 Spring 2012 E. Uysal-Biyikoglu
29
Definition 5 A discrete random variable takes a discrete set of values.
The Probability Mass Function (PMF) of a discrete random variable is
defined as
pX (x) = P(X = x)
Ex: Find and plot the PMFs of X and Y defined in the previous
example.
• A discrete random variable is completely characterized by its PMF.
Ex: Let M be the maximum of the two rolls of a fair die. Find pM (m)
for all m. (Think of the sample space description and the sets of outcomes
where M takes on the value m.)
METU EE230 Spring 2012 E. Uysal-Biyikoglu
2.2
30
Some Discrete Random Variables
2.2.1
The Bernoulli Random Variable
In the rest of this course, we shall define the Bernoulli random variable
with parameter p as the following:

1 with probability p
X=
0 with probability 1 − p
In shorthand we say X ∼Ber(p).
Ex: Express and sketch the PMF of a Bernoulli(p) random variable.
Despite its simplicity, the Bernoulli r.v. is very important since it can
model generic probabilistic situations with just two outcomes (often referred to as binary r.v.).
Examples:
• Indicator function: Consider the random variable X defined previously. X(w) = 1 if outcome w ∈ A, and X(w) = 0 otherwise. So,
X indicates whether the outcome is in set A or Ac . X, a Bernoulli
random variable, is sometimes called the “indicator function” of the
METU EE230 Spring 2012 E. Uysal-Biyikoglu
31
event A. This is sometimes denoted as X(w) = IA (w).
• Consider n tosses of a coin. Let Xi = 1 if the ith roll comes up H, and
Xi = 0 if it comes up T. Each of the Xi ’s are independent Bernoulli
random variables. The Xi’s, i = 1, 2, . . . are a sequence of independent
“Bernoulli Trials”.
• Let Z be the total number of successes in n independent Bernoulli trials. Express Z in terms of n independent Bernoulli random variables.
2.2.2
The Geometric Random Variable
Consider a sequence of independent Bernoulli trials where the probability of success in each trial is p (We will later call this a “Bernoulli Process”.)
Let Y be the number of trials up to and including the first success. Y is a
Geometric random variable with parameter p.
P(Y = k) =
for k =
METU EE230 Spring 2012 E. Uysal-Biyikoglu
32
Sketch pY (k) for all k.
Check that this is a legitimate PMF.
Ex: Let Z be the number of trials up to (but not including) the first
success. Find and sketch pZ (z).
2.2.3
The Binomial Random Variable
Consider n independent Bernoulli Trials each with probability of success
p, and let B be the number of successes in the n trials. B is Binomial with
parameters (n, p).
33
METU EE230 Spring 2012 E. Uysal-Biyikoglu
P(B = k) =
for k =
Ex: Let R be the number of Heads in n independent tosses of a coin
with bias p.
n=7, p=0.25
0.4
X
p (k)
0.3
0.2
0.1
0
0
1
2
3
4
5
6
7
k
n=40, p=0.25
0.2
pX(k)
0.15
0.1
0.05
0
−5
0
5
10
15
20
k
25
30
35
40
45
METU EE230 Spring 2012 E. Uysal-Biyikoglu
2.2.4
34
The Poisson Random Variable
A Poisson random variable X with parameter λ has the PMF
λk e−λ
pX (k) =
, k = 0, 1, 2, . . .
k!
Ex: Show that
eλ .
P
k
pX (k) = 1 (Hint: use the Taylor series expansion of
• The Binomial is a good approximation for the Poisson with λ = np
when n is very large and p is small, for small values of k. That is, if
k≪n
2.2.5
n!
λk e−λ
≈
pk (1 − p)(n−k)
k!
k!(n − k)!
The Discrete Uniform R.V.
The discrete uniform random variable takes consecutive integer values
within a finite range with equal probability. That is, X is Discrete Uniform
in [a, b], b > a if and only if
pX (k) = 1/(b − a + 1) for k = a, a + 1, a + 2, . . . , b
Ex: A four-sided die is rolled. Let X be equal to the outcome, Y be
equal to the outcome divided by three, and Z be equal to the square of the
outcome.
35
METU EE230 Spring 2012 E. Uysal-Biyikoglu
(Note that Y and Z both take four equally likely values, however they do
not have the discrete uniform distribution.)
2.3
Functions of Random Variables
Y = f (X)
Ex: Let X be the temperature in Celsius, and Y be the temperature
in Fahrenheit. Clearly, Y can be obtained if you know X.
Y = 1.8X + 32
Ex: P(Y ≥ 14) = P(X ≥?)
Ex: A uniform r.v. X whose range is the integers in [−2, 2]. It is passed
through a transformation Y = |X|.
To obtain pY (y) for any y, we add the probabilities of the values x that
results in g(x) = y:
pY (y) =
X
pX (x).
x:g(x)=y
Ex: A uniform r.v. whose range is the integers in [−3, 3]. It is passed
METU EE230 Spring 2012 E. Uysal-Biyikoglu
36
through a transformation Y = u(X) where u(·) is the discrete unit step
function.
2.4
Expectation, Mean, and Variance
We are sometimes interested in a summary of certain properties of a
random variable.
Ex: Instead of comparing your grade with each of the other grades in
class, as a first approximation you could compare it with the class average.
Ex: A fair die is thrown in a casino. If 1 or 2 shows, the casino will pay
you a net amount of 30, 000 TL (so they will give you your money back
plus 30,000), if 3, 4, 5 or 6 shows you they will take the money you put
down. Up to how much would you pay to play this game?
Ex: Alternatively, suppose they give you a total of 30, 000 if you win
(regardless of how much you put down), and nothing if you lose. How
much would you pay to play this game?
(answer: the value of the first game (the break-even point) is 15,000,
and for the second game, it is 10,000. In the second game, you expect to
get 30,000 with probability 1/3, so you expect to get 10,000 on average.)